braedonsaunders/bunkhouse ? reverse-engineered prompt
Reverse engineered prompt
Build me an open source web app for running AI employees inside a small business.
I want a place where each agent has a real company email address, a role, working hours, memory, and a budget, and coworkers can just email it like a normal employee. The main screen should feel like one conversation, with chat, calls, approvals, and a live work area all together, so I can watch what the agent is doing in its browser, terminal, files, desktop, and history without jumping around.
Make it support human approvals for risky actions, keep a clear audit trail of messages, tool use, and outcomes, and let people attach files and get real documents back like PDFs, Word files, and spreadsheets. It should be able to connect to Google Workspace, Microsoft 365, or regular IMAP and SMTP mail, and it should work in Docker with a simple local setup. If you need to look up current docs online while building it, go ahead.
Are you gonna build this?
make sure you review the code using arcumet